11

Chapter 2

Seismic Response of Rail-Counterweight

System

This chapter focuses on the analytical study of a rail-counterweight system exposed to seismic

motions. Previous analytical and experimental studies done on this system are reviewed. The

deficiencies in the models used in these studies are noted. The rest of the chapter is then

devoted to describe the development of the analytical model used in this study including the

effect of system nonlinearities, and then evaluate the seismic performance of the system under

several recorded and synthetically generated accelerograms. The performance of the system is

evaluated in terms of overstressing of the rails and bracket supports, and also in terms of the

system fragility for an ensemble of seismic motions.

2.1 Previous Studies on Rail-Counterweight in Elevators

The observed seismic damages on elevators, especially the rail-counterweight system, have

been followed by a continued research interest on the seismic response characteristics and

performance of elevators in buildings. As most of the damage happened to the rail-

12

counterweight systems, most of the studies were focused on the dynamic behavior of the rail-

counterweight system.

Yang et al. (1983) studied the nonlinear response of physical model of counterweight

subjected to harmonic excitation in the in-plane direction and compared the results with those

of finite element method. The counterweight model consisted of two end plates connected by

a central member and two flexible members representing the frame, and five small shafts

attached perpendicular to the central member representing the weights. The weights were free

to slide within the counterweight frame. The counterweight was supported by a leaf spring at

each of its corner, simulating the roller guide assemblies. The rails were represented by two

steel bars with rectangular cross sections. In the finite element model, the rails were modeled

as simply supported beams and the counterweight frame was modeled as beam on elastic

supports. The springs that connected the counterweight and the rail were assumed to behave

as bilinear springs, which had zero values when there was no contact and very large value

when they were in contact. Because of difference in the physical characteristics between the

two methods, only the general forms of the responses were compared.

Tzou and Schiff (1984, 1989) used similar model to evaluate the dynamic loading due to

contact between the weights and the frame. The rail and the counterweight frame were

modeled as a single combined beam. The counterweight was assumed to be located in the

middle of the rail span. A large separation was provided between the weight and the frame. It

was found that the seismic response of the counterweight can be improved by providing a

large separation so that contact was avoided. However, when contact did happen, the response

was worse than that of the regular small gap. One modification that was suggested if contact

occurred was installation of rubber damper between the frame and the weights. It was

13

concluded that the rubber dampers could decrease the rail load provided that proper values of

coefficients were selected.

Segal et al. (Segal et al. 1994, 1995, 1996; Rutenberg et al. 1996; Levy et al. 1996)

studied the seismic response of counterweight modeled as two independent mass located at

the upper and lower guiding members. The counterweight was guided along the rail by sliding

shoes. Therefore, it hanged freely before contact happened. Contact elements were used to

represent rail-brackets system, using parallel combination of spring, dashpot, and Coulomb

friction element. The vertical velocity of the counterweight was included in the analysis.

Because of the Coulomb friction element, which would come into the equations when contact

occurred, different vertical velocity would result in different equivalent stiffness of the rail.

An ensemble of recorded earthquake ground motion normalized to 0.4 m/s and 0.3 m/s peak

ground velocities were used an input motion to the building. All bracket supports were

assumed to be subjected to the same acceleration equal to that at the top of the building.

In all these previous studies, not much attention was given to the proper modeling of the

rail-counterweight systems, especially of their roller guide assemblies. In some respect their

models were different from each other. Suarez and Singh (1996, 1998) were perhaps the first

one to include the details of the roller guide assembly in the preparation of their analytical

model. To study the out-of-plane motion of the counterweight, they used three-degrees-of-

freedom model which included equivalent springs representing the roller guide assemblies,

rails, and bracket supports. The continuous beam effect of the rail was also included. In the

first study, the model was reduced to a single degree-of-freedom system by relating the

frequencies of the rotation modes to the translation mode, and response spectrum approach

was used to calculate the response. In the second paper, the effect of saturation of the roller

14

guide was included. This model also ignored several important features of the counterweight

systems used in elevators, for example, the code limits on the clearances between the

restraining plates at the roller guide assemblies and clearance between the counterweight

frame and the rail.

2.2 Analytical Model for Rail-Counterweight System

In this study, the counterweight frame is modeled as a rigid block of height lc, width d, and

depth e. The weights fill the bottom two-third of the frame and the center of mass of the

counterweight is located at a distance lm from the bottom of the frame, as shown in figure 2.1.

The flexibility of the system is provided by the combination of brackets, guide rails, and roller

guide assemblies, represented by an equivalent spring at each of the four corners of the

counterweight frame. The stiffness of the supporting system of the counterweight is obtained

by taking into account the continuous rails on bracket supports and the stiffness of the roller

guide assembly. The values of the combined stiffness could be different for the upper and

lower roller guides depending upon their locations on the guide rails.

The dynamic motions of the counterweight under seismic load can be described in five

degrees of freedom, two translations u and v, and three rotations ψ, θ, and φ, as also shown in

figure 2.1. The degrees of freedom u and ψ are associated with in the in-plane motion of the

counterweight while others are associated with the out-of-plane motion. The equations of

motion for these two directions are developed separately in the following sections.

2.2.1 Sources of Nonlinearities

There are several sources that can introduce nonlinearity to the rail-counterweight system

during its vibrations. The first source is the force-deformation characteristics of the tires of the

roller guide. The nonlinear behavior of the tires can be established in a reliable manner only

15

through experiments. However, as shown by Diaz (1999), within the range of deformation

during earthquake-induced vibration the force-deformation relationship can be assumed to be

linear.

Yielding of the rails or brackets due to excessive stress can also cause nonlinearity in the

system. However, the post-yielding behavior of the rail-counterweight system is not much of

practical interest since the damage would have been permanent. In fact, the main objective of

the study is to avoid this failure.

Saturation of the helical spring in the roller guide assembly when the deformation close

the gaps between helical wire loops can also introduce nonlinearity to the system. This can be

avoided by proper design of the spring. Moreover, if this saturation occurs before the contact

between restraining plate and the rail it can be included in the analysis by modifying the

force-deformation characteristics associated with the contact.

The sources of nonlinearity that are included in this analysis are associated with closing

of the gap between the restraining plated and the rails at any of the four roller guide

assemblies and between the counterweight frame and the rail at the bracket support. As

mentioned earlier in Chapter 1, a restraining plate is provided at each roller guide assembly to

avoid roller guide coming off the guide rails during earthquake. The code requires a

maximum clearance of 3/16 inches between the restraining plate and the rail in both in-plane

and out-of-plane directions. This gap is shown in Figure 1.4 in Chapter 1. During the

earthquake-induced vibration, this gap could possibly be closed especially under medium or

strong ground motion. When contact between one of the restraining plates and the rail

happens, the flexibility will be provided only by the guide rail and the supporting bracket.

Because the springs representing the roller guide assemblies and rail-brackets are in series

16

with each other, and the guide rails and brackets are much more rigid than the helical springs

and rubber tires, the stiffness of the system will increase significantly after the contact.

Therefore, the force-deformation relationship of the equivalent spring at each corner can be

represented by bilinear diagram as shown in Figure 2.2.

The middle, almost flat, part of the diagram represents the equivalent spring before

contact happens when the tangent stiffness kt is governed by the roller guide assembly. On the

steep portion of the diagram, the tangent stiffness kt� is governed by that of the guide rail and

bracket. The relationship between the force f and deformation δ can then be written as

( )sgnmf k kδ δ δ= − ∆ (2.1)

where

'm

if if >

t m

t

kk

kδ δδ δ

≤=

(2.2)

't tk k k∆ = − (2.3)

δm is the maximum deformation before contact happens, and sgn is the sign function that

returns the value of �1 if δ < 0, 0 if δ = 0, or 1 if δ > 0.

During the in-plane motion of the rail-counterweight system, one of each pair of rollers

at the upper and lower roller guide assemblies will always be in contact with the rail,

providing the restoring forces fu and fl. Using (2.1), these forces can be written as

( )( )

sgn0sgn0

iu uu iu um

il ll il l

k uf k uu

k uf k u∆ = − ∆

(2.4)

where uu and ul is the in-plane displacement of the equivalent spring at the upper and lower

supports of the frame, respectively, and um is the maximum displacement before contact

happens. Unless otherwise noted, subscripts u and l will be used throughout this thesis to

17

respectively denote the upper and lower part of the counterweight frame. Similarly, subscripts

i and o will denote the values in the in-plane and out-of-plane directions, respectively.

For the out-of-plane direction, all equivalent spring at the four corners of the frame

contribute in providing the elastic force as follows

( )( )( )( )

1 11 1 1

2 22 2 2

1 11 1 1

1 22 2 2

sgn0 0 0sgn0 0 0sgn0 0 0sgn0 0 0

ou uu ou u

ou uu ou um

ol ll ol l

ol ll ol l

k vf k vk vf k v

vk vf k vk vf k v

∆ ∆ = − ∆ ∆

(2.5)

The code also limits the gap between the counterweight and the frame to be not more

than half of an inch. If the combined deformations of the guide rail and the roller guides at the

two ends are large enough to overcome this gap, the bracket support will provide another

resisting force for the in-plane motion of the rail-counterweight system. The force-

deformation diagram for this case is shown in figure 2.3.

This can happen only when the upper and lower roller guides are located on the upper

and lower sides of a bracket support, respectively. In other words, the position of the

counterweight is such that

u ca L l> − (2.6)

where au is the distance from the upper roller guide to the next bracket support above it, L is

the story height (or distance between two consecutive bracket supports), and lc is the height of

the counterweight.

The force fs can be calculated from

( )( )sgns br s sm sf k u u u= − (2.7)

18

where us is the in-plane displacement of the counterweight at the bracket support level, usm is

the maximum displacement before contact happens, and

if and

0 otherwise br inplane u c s sm

brk a L l u u

k − > − >=

(2.8)

represents the tangent stiffness of the force-deformation diagram that is provided by the

bracket.

Combining (2.4) and (2.7), the restoring force vector in the in-plane direction becomes

( )( )( )

0 0 sgn0 0 sgn0 0 sgn

u iu u m iu u

l il l m il l

s br s sm br s

f k u u k uf k u u k uf k u u k u

∆ = − ∆

(2.9)

2.2.2 In-Plane Motion

Figure 2.4 shows the three forces and the inertial forces acting in the in-plane direction. Using

the balance of force along the x-axis and moment about the y-axis, we obtain

( ) ( )u l s c c cf f f m u x− + + = +!! !! (2.10)

( ) ( ) ( )u c m s c m s l mf l l f l l l f l Jψ ψ δ− − − − − + = +!!!! (2.11)

where ls is the distance between the upper roller guide to the contact location at the bracket

support level

s ul L a= − (2.12)

and

2 2

213 4c m

m

m l dJlψ

= +

(2.13)

Introducing new variables

19

e mu l ψ= (2.14)

( )me m u l

c

lx l x xl

δ= = −!!!! !! !! (2.15)

equations (2.10) and (2.11) can be written in matrix form

+ = −i i i iM q F M x!! !! (2.16)

where

1

1 0 ; ;

0c

ce e

u xm

u xγ

= = =

i iM q x!!

!!!!

(2.17)

1 1 1

1 1 1

u u

l lc c s

m m m s s

f ff fl l l

l l l f f

= = − − − −

Ti iF T (2.18)

2

1 2

1 13 4 m

dl

γ

= +

(2.19)

The displacement vector in (2.9), which contains the displacement at the upper and

lower roller guide as well as at the possible contact location between the frame and the rail,

can be related to the degrees of freedom of the system qi with transformation matrix Ti as

follows

1 1

1 1

1 1

c

muc

le

s c s

m m

llu

uu

uu l l

l l

− = − = − −

i iT q (2.20)

Substituting equations (2.20) and (2.9) into (2.18), the elastic force can be expressed in terms

of the tangent stiffness matrix and the nonlinear force component

= −i it i iF K q f (2.21)

20

where

0 0

0 00 0

iu

il

br

kk

k

=

Tit i iK T T (2.22)

( )( )( )

sgnsgnsgn

m iu u

m il l

sm br s

u k uu k uu k u

∆ = ∆

Ti if T (2.23)

The equations of motion (2.16) can now be written as

+ = − +i i it i i iM q K q M x f!! !! (2.24)

2.2.3 Out-of-plane Motion

The free-body diagram for the out-of-plane motion of the counterweight is shown in figure

2.5. As mentioned earlier, there are four resisting forces at each corner of the frame. The three

equations of motion are obtained by summing up forces in the y-direction perpendicular to the

plane of the counterweight, and by balancing the moments about the z- and x-axes

( )1 2 1 2u u l l c c cf f f f m v y+ + + = − +!! !! (2.25)

( )( ) ( ) ( )1 2 1 2u u c m l l mf f l l f f l Jθ θ β− − + − + + = + !! !! (2.26)

( ) ( )1 2 1 2 2u u l ldf f f f Jφ φ α− − + − = +!! !! (2.27)

where

2 2

2112cm d eJ

dθ

= +

(2.28)

2 2

213 4c m

m

m l eJlφ

= +

(2.29)

21

Following similar procedure as the in-plane motion analysis, we introduce the following

response variables

; e m e mv l y lθ β= = !!!! (2.30)

, , ; 2 2e ed dv yθ α= = !!!! (2.31)

and write the equations of motion in matrix form

+ = −!! !!o o o oM q F M y (2.32)

where

2, ,

3

1 0 00 0 ; ; 0 0

c c

c e e

e e

v ym v y

v yγ

γ

= = =

!!!! !!

!!o oM q y (2.33)

1 1

2 2

1 1

2 2

1 1 1 1

1 1 1 1

1 1 1 1

u u

u uc c

l lm m

l l

f ff fl lf fl lf f

= − − = − −

To oF T (2.34)

2

2 2

1 13 4 m

el

γ

= +

(2.35)

2

3 2

1 13 m

ed

γ

= +

(2.36)

The displacements of the four equivalent springs can also be related to the displacement

vector qo using transformation matrix To

1

2

1

2

u

u

l

l

vvvv

=

o oT q (2.37)

22

Combining (2.34), (2.5), and (2.37), the vector force Fo can then be written in terms of

tangent stiffness matrix Kot

= −o ot o oF K q f (2.38)

where

1

2

1

2

0 0 00 0 00 0 00 0 0

ou

ouo o o

ol

ol

kk

kk

=

TtK T T (2.39)

( )( )( )( )

1 1

2 2

1 1

2 2

sgnsgnsgnsgn

ou u

ou uo m o

ol l

ol l

k vk v

vk vk v

∆ ∆ = ∆ ∆

Tf T (2.40)

Using (2.38), the equations of motion can then be written as

+ = − +o o ot o o oM q K q M y f!! !! (2.41)

In the numerical integration, the equations of motion for the in-plane motion and for out-of-

plane motion are combined into

+ = − +tMq K q MY f!!!! (2.42)

where

;

= =

i itt

o ot

M 0 K 0M K

0 M 0 K (2.43)

=

i

o

qq

q (2.44)

=

xY

y!!!!!!

(2.45)

23

2.2.4 Damping Mechanism

The equations of motion in the previous sections were derived without any damping

mechanism in the system. In this study, a viscous damping matrix is added to incorporate

some inevitable energy dissipation in the system. The damping matrix is defined in terms of

constant modal damping ratio as

2ζ= TjC MΦω Φ M (2.46)

where ζ is the modal damping ratio, Φ is the matrix of eigenvectors, and ωj is the diagonal

matrix of the system frequencies. The final form of the equations of motion with the damping

matrix included becomes

+ + = − +tMq Cq K q MY f!!!! ! (2.47)

2.2.5 Input Motion

The acceleration inputs in the equations of motion come from the ground acceleration that is

filtered by the building structure. Therefore, the input acceleration vector Y!! can be expressed

in terms of the absolute accelerations of the building floors. For a torsional building, each

floor has three components of accelerations: translation in two perpendicular directions

and n nU V!! !! , and rotation nΘ!! . When the upper roller guides are located between the n-th and

(n-1)-th floors, the accelerations at the roller guides position can be obtained from linear

interpolation of accelerations of both floors

( ) ( ) ( )1 1 1 11 u uu n n n n

a ax t U y U yL L − −

= − − Θ + − Θ

!! !! !! !!!! (2.48)

( ) ( ) ( )1 1 1 1 11 u uu n n n n

a ay t V x V xL L − −

= − + Θ + + Θ

!! !! !! !!!! (2.49)

24

( ) ( ) ( )2 2 1 2 11 u uu n n n n

a ay t V x V xL L − −

= − + Θ + + Θ

!! !! !! !!!! (2.50)

where x1, x2, and y1 are the distances from the center of mass of the floor to the guide rails as

shown in figure 2.6.

The lower roller guides can be located on the same span of rail or on the next span

below. When the counterweight is located within a single span of the rail, the accelerations at

the lower roller guides are calculated from

( ) ( ) ( )1 1 1 11 l ll n n n n

a ax t U y U yL L

Θ Θ− − = − − + −

!! !! !! !!!! (2.51)

( ) ( ) ( )1 1 1 1 11 l ll n n n n

a ay t V x V xL L

Θ Θ− − = − + + +

!! !! !! !!!! (2.52)

( ) ( ) ( )2 2 1 2 11 l ll n n n n

a ay t V x V xL L

Θ Θ− − = − + + +

!! !! !! !!!! (2.53)

If the counterweight is located on two consecutive spans, the lower roller guides will be

between the (n-1)-th and (n-2)-th floors. In this case the accelerations at the roller guide level

becomes

( ) ( ) ( )1 1 1 2 1 21 2l ll n n n n

a ax t U y U yL L− − − −

= − − Θ + − − Θ

!! !! !! !!!! (2.54)

( ) ( ) ( )1 1 1 1 2 1 21 2l ll n n n n

a ay t V x V xL L

Θ Θ− − − − = − + + − +

!! !! !! !!!! (2.55)

( ) ( ) ( )2 1 2 1 2 2 21 2l ll n n n n

a ay t V x V xL L

Θ Θ− − − − = − + + − +

!! !! !! !!!! (2.56)

where al = au + lc.

Finally, the input acceleration terms in (2.17) and (2.33) can be obtained from

25

( ) ( )mc l u l

c

lx t x x xl

= + −!! !! !! !! (2.57)

( ) ( )me u l

c

lx t x xl

= −!! !! !! (2.58)

( ) ( )1 2 1 212

m cc u u l l

c m

l ly t y y y yl l

= + + − +

!! !! !! !! !! (2.59)

( ) ( )1 1m

e l uc

ly t y yl

= −!! !! !! (2.60)

( ) ( ),1 2

12e u uy t y y= −!! !! !! (2.61)

2.3 Numerical Results

Physical Properties of the System

A typical 4300-lb counterweight located in a 10-story torsional building is used in the

analysis. The building is a slight modification from the building described by Conner et al.

(1987). The building has a rectangular plan with a small eccentricity. As mentioned earlier,

each floor of the building has two translational and one rotational degree-of-freedoms. The

mass and stiffness properties of the building are shown in Table 2.1. The natural frequencies

and mass matrix normalized participation factors are presented in Table 2.2. The building is

assumed to have 5% modal damping ratio for all modes.

The 18.5-lb guide rail with no intermediate tie brackets is used as the basic system in the

numerical analysis. The cross-sectional properties of this guide rail are shown in Table 2.3

along with other sizes provided by the code. The bracket supports are located on each floor

level with 12 ft story height. The dimensions of the counterweight are height lc = 138 in.,

26

width d = 28 in., and depth e = 6 in. The helical springs in the roller guide assemblies are

made of steel with five coils having mean radius of 0.5 in and wire diameter of 0.25 in. The

clearances between the restraining plates and the rail and between the frame and the rail are

set to the maximum allowed by the code. An inherent modal damping ratio of 2% is assumed

for all modes.

Seismic Inputs

Six sets of recorded ground accelerations, listed in Table 2.4, are used to excite the building to

provide seismic input to the rail-counterweight system. Each set of accelerogram consists of

two horizontal components in the perpendicular directions. In addition, a set artificial

accelerogram corresponding to broadband Kanai-Tajimi spectral density function are also

used as input to the building. For each intensity level, 50 sets of accelerograms are generated

for this purpose. The average and mean plus one standard deviation pseudo-acceleration

spectra of the sets with 0.1g maximum ground acceleration are shown in Figure 2.7.

The spectral density is in the form of

4 2 2 2

2 2 2 2 2 2

4 ( )

( ) 4 g g g

g g g

Sω + β ω ω

Φ ω =ω − ω + β ω ω

with parameters ωg = 23.96 rad/s and βg = 0.32. The synthetic ground acceleration f(t) is

generated as the sum of k harmonic functions with frequencies ωk and random phase angles

δk,

{ }( ) ( ) 4 Re ( ) k ki i tk

kf t t e eδ ω = κ ∆ω Φ ω∑

where κ(t) is a deterministic envelope function:

27

( )

2

0.26 10

( / 3) 0 s 3 s( ) 1 3 s 10 s

10 st

t tt t

e t− −

≤ ≤

κ = ≤ ≤ >

Seismic Response

A unique characteristic of the rail-counterweight system is that the frequency of the system

change not only because of the contacts but also with different location of the counterweight

along the rail. Figure 2.8 shows the natural frequencies of the system as a function of

counterweight position along the building height. The frequencies shown are for the original

system without any contacts between counterweight and the rail. If contact happens, the

frequencies become much higher. For example, when the counterweight is at the top of the

building, the first natural frequencies of the in-plane motion increases from 12.3 rad/s to 23.7

rad/s if contact happens at the lower roller guide assembly.

Figure 2.9 shows the time history of the in-plane displacement of the lower part of the

counterweight under the actual El Centro ground motion. There are several occasions that the

displacement exceeds the gap between the restraining plate and the rail, which is represented

by the dashed line on the figure. It is also quite obvious that the frequency of the oscillation

increases when this gap closed. The contact happens only for a short period of time. This

requires the time step integration to be small enough to capture it. The fourth order Runge-

Kutta method with adaptive time step was used in this analysis for its numerical efficiency

and accuracy. For the sake of completeness, a numerical side-study performed in this

connection is described in Appendix A.

28

Figures 2.10 and 2.11 show the maximum stresses in the rail and brackets for different

counterweight positions along the building height. The maximum acceleration of the building

floor under Northridge and El Centro earthquakes with maximum ground acceleration

normalized to 0.1g are also plotted in the same figures. In both cases, the building actually

amplifies the maximum acceleration of the ground motion and the highest acceleration occurs

at the top of the building. However, the maximum stress in the rail or bracket does not

necessarily happen when the counterweight is located at the top of the building.

The maximum stress in the web of the rail is higher than the maximum stress in the

flange in both figures, indicating that the effect of the in-plane motion is more dominant than

the out-of-plane. The patterns of maximum stresses are qualitatively similar for counterweight

at different span of the rail. The stress in the bracket is higher when the lower roller guides are

located close to the bracket support while the stress in the rail is higher when the

counterweight is about at the middle of the rail span.

Parametric Study

The effects of several parameters of the rail-counterweight system are studied by conducting a

parametric study. Since the trends of stress and force responses at different span of the rail are

similar, the parametric study compares only the responses when the counterweight is located

on the top story. First we look at the effect of the clearances required by the code, both

between the restraining plate and the rail and between the frame and the rail. Other parameters

are the size of guide rail, different ground motions and their intensities, and the use of

intermediate tie-brackets.

Clearance at the Restraining Plate: Figures 2.12 and 2.13 show the maximum stresses in the

rail and force in the brackets, respectively, for different clearances between the restraining

29

plate and the rail. Results under 0.1 g Northridge and El Centro earthquakes are both

presented in the figures. The code limits this clearance to be no more than 3/16 of an inch, as

indicated by the dashed line in the figures. From this limit, it is clear that the maximum stress

and force increase if the gap is widened up to an optimal point, and then decrease until the gap

is wide enough that there are no contacts. On the other hand, reducing the gap size seems to

improve the responses. However, this also means more contact and probably more impact

noise during the earthquake.

Clearance Between the Frame and Rails: The effect of clearance between the frame and the

rail is presented in figures 2.14 to 2.16. The code requires that this clearance to be 0.5 inch or

less in order to limit the rail deformation in the in-plane direction when the counterweight is

located on two consecutive spans. If the deformation of the rail closes the gap, part of the

force will then be transferred directly to the support. The results shown in Figures 2.14 to 2.16

are obtained with actual Northridge (0.843g) and El Centro (0.348g) earthquakes because the

low intensity of 0.1g considered earlier did not produce any contact between the rail and the

frame.

Figure 2.14 shows that the maximum stress in the rail can be reduced significantly,

especially for Northridge case, if the clearance is reduced. There are no significant effects on

the maximum stress under El Centro earthquake, but the maximum stress in this case is

already low. Figure 2.15 shows similar plot for the maximum in-plane force in the bracket.

The maximum in-plane force seems to increase with decreasing clearance for El Centro

earthquake but this is not always the case for the Northridge earthquake. This maximum in-

plane force under Northridge earthquake is then plotted in Figure 2.16 as a function of

counterweight location on the top story of the building for different values of clearances. This

30

figure shows that for small clearance, the contacts occur at a large range of counterweight

position in the middle of rail span. For larger clearance, the range is more confined to the

counterweight location near the middle of the span. However, this figure indicates that larger

range of contact does not necessarily produce higher force in the bracket.

Effect of Rail Size: The next set of figures present the response of the rail-counterweight

system for different rail sizes, 12-lb, 18.5-lb, and 30-lb. The maximum stress in the rail under

actual Northridge and El Centro earthquakes are plotted respectively in Figures 2.17 and 2.18,

while Figure 2.19 shows the maximum in-plane force under actual Northridge earthquake. As

expected, larger rails experience smaller maximum stress. In addition, smaller rails have

larger deformation that creates wider range of contact in the middle of the span, as shown in

Figure 2.19, and thus produces higher in-plane force in the bracket. It should be noted that

according to the code all the three sizes are acceptable for the 12 ft. support spacing without

intermediate tie-bracket that is used in this study. However, Figure 2.17 shows that even the

18.5-lb. rail is not sufficient for actual Northridge earthquake.

Effect of Different Seismic Inputs: Figures 2.20 to 2.26 show the effects of different ground

motions and different intensities to the responses of the rail-counterweight system. The six

earthquake records mentioned in table 1 are used as the base motion to the building. In Figure

2.20, the maximum stresses in the rail under the actual earthquakes, which of course have

different intensities. This figure shows that the stress under Northridge earthquakes is the

highest although the maximum ground acceleration of the earthquake is not the highest. In

Figure 2.21, all earthquakes are normalized to 0.5g. Again, Northridge earthquake produces

the highest stress in the rail. The effect of actual El Centro is below average of the actual

earthquakes but it becomes close to the effect of Northridge when all earthquakes are

31

normalized to 0.5g. Similar observation can be obtained from figure 2.22 for the in-plane

force in the bracket. The normalized Northridge earthquake gives the highest force followed

closely by El Centro. Figures 2.23 to 2.25 show the average, maximum, and minimum values

of the stresses and forces in Figures 2.20 to 2.23. Comparing Figures 2.23 and 2.24, it can be

seen that normalizing the earthquake inputs to the same maximum ground acceleration will

narrow the range of the maximum stress in the rail. Finally, in Figure 2.26 the maximum

stresses in the rail are plotted for increasing maximum ground acceleration of the six

earthquake records. This figure show that a moderate Northridge or strong El Centro

earthquakes make the rail overstressed above the allowable stress while other earthquakes do

not have that effect. These point to the need of considering several ground motions that are

representative to the site of the building in evaluating the performance of the rail-

counterweight system of an elevator.

Effect of Tie-Brackets: Figures 2.27 to 2.30 show the effect of installation of intermediate

tie-bracket to connect the mid-span of the rail. According to the code, larger spacing of

brackets is allowed if the tie-brackets are installed to tie the rails. In this study the tie-brackets

are added to the original system without changing the bracket spacing. Figures 2.27 and 2.28

show that the intermediate tie-bracket can reduce the maximum stress in the rail when the

counterweight is located at the middle of the rail span. It also reduces the contact between the

frame and the rail, and thus reducing the force due to this contact, as shown in Figures 2.29

and 2.30. However, all these figures also show that the stress and force when the

counterweight is at other locations, especially near the bracket support, could increase. In fact,

for low intensity of Northridge (0.1g) the stress in the rail with tie-bracket is higher than the

original system.

32

Fragility Study

Up to this point, the responses of the rail-counterweight have been obtained with several

recorded ground motions from different events at different locations. The next set of results

presents the statistics of the responses under an ensemble of ground accelerations having

similar frequency characteristics as mentioned earlier in this section. Besides comparing the

mean values or mean plus one standard deviation values of the responses to their maximum

allowable values, the results can also be presented in terms of fragility of the rail-

counterweight system.

Fragility of a system or its component is defined as the conditional probability of failure

for a given value of seismic intensity parameter such as peak ground acceleration. It provides

an important piece of information needed for seismic risk analysis of structural or mechanical

systems (Kennedy and Ravindra, 1984). For the rail-counterweight system considered in this

study, the failure is defined in terms of the induced stress S in the rail or bracket exceeding the

yield strength R of the material. For a given maximum ground acceleration y, the fragility

expressed as the conditional probability of failure can be written as

( )F y P S R Y y= ≥ = (2.62)

To estimate this probability of failure, the probability distributions of R and S must be

known. Here, it is assumed that R and S are lognormally distributed. For the yield strength R

of the rail and bracket material, the mean value equal to 1.05 times the nominal value with the

coefficient of variation of 0.1 is assumed (Brockenbrough, 1999). Similar parameters for the

induced stress S are obtained by dynamic analysis of the rail counterweight system for the

ensemble of ground acceleration mentioned above.

33

For each counterweight position along the rail, failure can happen at different locations

of the rail. The probability of failure for a specific location of the counterweight, Pf-i, is

determined by the largest of the probabilities of failure Pf-ij of seven different points on each

rail and six bracket locations affected by each counterweight position

maxf i f ijjP P− −= (2.63)

To obtain the probability of failure considering all counterweight positions, the

probability of failure for each location of counterweight is then convoluted with the

probability of counterweight being on that specific location. Assuming that the counterweight

can be anywhere along the rail with equal probability, a uniform distribution for

counterweight position is justified. With this the unconditional probability of failure of the

system, considering all positions of the counterweight is equal to the average the probabilities

of failure calculated at discrete locations along the height

1f f i

iP P

N −= ∑ (2.64)

where N is the number of discrete locations where the probabilities of failure were calculated.

Figure 2.31 shows the statistics of the peak rail stress under the ensemble of synthetic

ground motions. Although the mean values are still below maximum allowable stress up to

1g, the mean plus one standard deviation and the absolute maximum values present the

possibility of overstress in the rail. The figure also shows that the increase in the responses

does not vary linearly with the increase of maximum ground acceleration. Figures 2.32 and

2.33 compare the responses of the system with and without intermediate tie-bracket. The tie-

bracket seems to help in reducing the peak stress and the fragility of the rail especially for

strong earthquakes.

34

2.4 Concluding Remarks

An analytical model of the rail-counterweight system has been developed incorporating the

detail of the guiding and support system of the counterweight, and the system nonlinearities.

Nonlinearity caused by closing and opening of the clearances at the restraining plates is

included in the model by using equivalent springs with bilinear force-deformation relationship

at the four corners of the counterweights. Similarly the nonlinearity caused by the on-off

contact of the frame with the rails at the bracket supports is included through the use of a

spring with another bilinear models added at the point of contact of the frame with the rails.

Depending upon the position of the counterweight along the building height, the location of

this spring on the frame would change. The model also includes the effect of differential

support motion filtered through the building and applied at the bracket supports.

The analysis shows that although the building accelerations are usually heights in the

top story, the maximum response in the counterweight-rail system does not necessarily occur

there. The patterns of the variation of the maximum stress in different spans of the rails in

different stories responses are similar. The stresses in the rails are higher when the top roller

guide is at about the middle of the rail span, while the highest force in the bracket happens

when the roller guide is on or near a bracket support.

The maximum stress in the rail and force in the bracket can be reduced by reducing the

3/16 inch gap specified by the code. However, this will cause more contact between the

restraining plates and the rail and thus more impact noise during earthquake. Similarly,

reducing the ½ inch gap between the frame and the rail can help reducing the maximum stress

in the rail, especially under strong earthquakes. The reduction will create larger range of

35

contact between the frame and the rail, but this larger range does not necessarily mean higher

force in the bracket.

The larger rail sizes, as expected, experience lower maximum stress and also have less

contact with the counterweight frame. The installment of intermediate-tie bracket can also

help in reducing the maximum stress in the rail, especially for strong earthquakes. However,

there is possibility that the stress will increase when the counterweight is located near the

support. This increase should be checked whether it would exceed the original maximum

stress, thus nullifying the effect of the tie-bracket.

The variation in the responses due to different ground motions is quite large. This points

to the need of considering an ensemble of site-specific ground motions to evaluate the

performance of the rail-counterweight system. The range of variation can be slightly reduced

if the maximum ground acceleration is normalized to a certain value. The maximum stress

also increases with increasing intensity of the ground motion. This increase in the stress,

however, does not relate linearly with the increase in the ground motion due to the

nonlinearities in the system.

Fragility of the rail-counterweight system expressed as the conditional probability of

failure of the rail, conditioned on the level of ground motion intensity, is also evaluated. In

later chapters, such fragility analyses will be used to measure the effectiveness of different

protective systems that are proposed for mitigating the seismic effects on the counterweight.

36

XY

Z

e

Ll

lu v

θ ψ

d

φ

c

m

rail

bracket

counterweight

rollerguide

ropes

counterweightframe

floor

Figure 2.1 Geometry of the five degree-of-freedom model of the rail-counterweight system.

37

f

δδδδm -δδδδm

δδδδ

Figure 2.2 Bilinear force-deformation diagram due to contact between the restraining plate and the rail.

fs

usm -usm us

Figure 2.3 Bilinear force-deformation diagram due to contact between the counterweight frame and the rail.

38

φJ (ψ + δ)

m (u + x )cc c

fu

fs

ψ

cueufl

Figure 2.4 Free body and kinetic diagram for the in-plane motion of the counterweight.

m (v + y )

z

d

J

uf 2

c c c

lc

v'

v

cvlv1

e

2

φ

θ

θ(φ + β)

(φ + α)Jφ

e

v 1uvc

lm

2d

2lf1lf

1uf

x

lm

Figure 2.5 Free body and kinetic diagram for the out-of-plane motion of the counterweight.

39

G

y

x

1

x12

X

Y

Figure 2.6 Location of the elevator hoistway on a typical floor plan.

40

0.1 1 10 1000.01

0.1

1

Frequency [Hz]

Pse

udo−

acce

lera

tion

Spe

ctra

[g]

mean mean + std. dev.

Figure 2.7 Mean and mean-plus-one-standard deviation of pseudo acceleration spectra of 50 sets of synthetic earthquakes with maximum ground acceleration of 0.1g

41

0 1 2 3 4 5 6 7 8 911.5

12

12.5

ω1 [r

ad/s

ec.]

0 1 2 3 4 5 6 7 8 933

33.5

34

34.5

ω2 [r

ad/s

ec.]

Ratio : au / L

(a) In-plane motion

0 1 2 3 4 5 6 7 8 916.5

17

17.5

ω1 [r

ad/s

ec.]

0 1 2 3 4 5 6 7 8 929.5

30

30.5

31

31.5

ω2 [r

ad/s

ec.]

0 1 2 3 4 5 6 7 8 948

49

50

51

ω3 [r

ad/s

ec.]

Ratio : au / L

(b) Out-of-plane motion

Figure 2.8 Natural frequencies of the system for different counterweight positions along the 10-Story Building for 18.5-lb. rail

42

0 2 4 6 8 10 12−0.25

−0.2

−0.15

−0.1

−0.05

0

0.05

0.1

0.15

0.2

0.25D

ispl

acem

ent [

in.]

Time [sec.]

Figure 2.9 Time history of the in-plane displacement at the lower support of the counterweight

43

0 5 10 15 20 25 30 35 40 45 501

2

3

4

5

6

7

8

9

10

Stress [ksi], Acceleration[g/200]

Cou

nter

wei

ght P

ositi

on (

Flo

or N

umbe

r)

(1) (2)

(5)

(3)

(4)

Figure 2.10 Maximum stress in the (1) rail flange, (2) rail web, and (3) brackets; and maximum floor acceleration in (4) the x-direction and (5) y-direction under Northridge

earthquake 0.1g.

44

0 5 10 15 20 25 30 35 401

2

3

4

5

6

7

8

9

10

Stress [ksi], Acceleration[g/200]

Cou

nter

wei

ght P

ositi

on (

Flo

or N

umbe

r)

(1)

(3)

(4)

(5)

(2)

Figure 2.11 Maximum stress in the (1) rail flange, (2) rail web, and (3) brackets; and maximum floor acceleration in (4) the x-direction and (5) y-direction under El Centro

earthquake 0.1g.

45

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20

5

10

15

20

25

30

35

Clearance [inch.]

Str

ess

[ksi

]NorthridgeEl Centro

Figure 2.12 Maximum stress in the rail for different clearance at the restraining plate for 0.1g Northridge and El Centro earthquakes.

46

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20

10

20

30

40

50

60

Clearance [inch.]

For

ce [k

ips]

out−of−plane, Northridgein−plane, Northridge out−of−plane, El Centro in−plane, El Centro

Figure 2.13 Maximum force in the bracket for different clearance at the restraining plate for 0.1g Northridge and El Centro earthquakes.

47

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 125

30

35

40

45

50

55

60

Clearance [inch.]

Str

ess

[ksi

]

Northridge El Centro

Figure 2.14 Maximum stress in the rail as a function of frame clearance for actual Northridge (0.843g) and El Centro (0.348g) earthquakes.

48

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 130

40

50

60

70

80

90

Clearance [inch.]

For

ce [k

ips]

Northridge El Centro

Figure 2.15 Maximum in-plane force in the bracket as a function of frame clearance for actual Northridge (0.843g) and El Centro (0.348g) earthquakes.

49

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

10

20

30

40

50

60

70

80

Ratio au / L

For

ce [k

ips]

∞ 0.2" clearance0.5" clearance0.8" clearance

Figure 2.16 Maximum force in the bracket as a function of counterweight position on the top story for different frame clearance under actual Northridge (0.843g) earthquake.

50

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

20

30

40

50

60

70

80

Ratio au / L

Str

ess

[ksi

]12−lb. rail18−lb. rail30−lb. rail

Figure 2.17 Maximum stress in different sizes of rail under actual Northridge earthquake (0.843g)

51

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 15

10

15

20

25

30

35

40

45

50

Ratio au / L

Str

ess

[ksi

]

maximum allowable stress12−lb. rail18−lb. rail30−lb. rail

Figure 2.18 Maximum stress in different sizes of rail under actual El Centro earthquake (0.348g)

52

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

20

40

60

80

100

120

Ratio au / L

For

ce [k

ips]

12−lb. rail18−lb. rail30−lb. rail

Figure 2.19 Maximum in-plane force in the bracket for different rail size, actual Northridge earthquake (0.843g).

53

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

15

20

25

30

35

40

45

50

55

60

Ratio au/L

Str

ess

[ksi

]El Centro Loma Prieta Northridge Parkfield San FernandoWhittier

Figure 2.20 Maximum stress in the rail for different base inputs (actual as recorded ground motion)

54

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 15

10

15

20

25

30

35

40

45

Ratio au/L

Str

ess

[ksi

]

El Centro Loma Prieta Northridge Parkfield San FernandoWhittier

Figure 2.21 Maximum stress in the rail for different base inputs, all normalized to 0.5g.

55

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

10

20

30

40

50

60

Ratio au/L

For

ce [k

ips]

El Centro Loma Prieta Northridge Parkfield San FernandoWhittier

Figure 2.22 Maximum in-plane force in the bracket for different base inputs, all normalized to 0.5g.

56

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

10

20

30

40

50

60

Ratio au/L

Str

ess

[ksi

]maximumaverageminimum

Figure 2.23 Average, maximum, and minimum values of the stress shown in figure 2.20.

57

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

10

20

30

40

50

60

Ratio au/L

Str

ess

[ksi

]maximumaverageminimum

Figure 2.24 Average, maximum, and minimum values of the stress shown in figure 2.21.

58

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

10

20

30

40

50

60

Ratio au/L

For

ce [k

ips]

maximumaverageminimum

Figure 2.25 Average, maximum, and minimum values of the in-plane force shown in figure 2.22.

59

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

15

20

25

30

35

40

45

50

55

60

Max. Ground Acceleration [g]

Str

ess

[ksi

]

maximum allowable stress

El Centro Loma Prieta Northridge Parkfield San FernandoWhittier

Figure 2.26 Maximum stress in the rail as a function of maximum ground acceleration for different earthquake inputs.

60

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 120

25

30

35

40

45

50

55

60

Ratio au / L

Str

ess

[ksi

]

maximum allowable stress

with tie−bracket without tie−bracket

Figure 2.27 Maximum stress in the rail with and without intermediate tie-bracket for actual Northridge earthquake.

61

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

15

20

25

30

35

40

45

50

Ratio au / L

Str

ess

[ksi

]

maximum allowable stress

with tie−bracket without tie−bracket

Figure 2.28 Maximum stress in the rail with and without intermediate tie-bracket for actual El Centro earthquake.

62

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

10

20

30

40

50

60

70

80

Ratio au / L

For

ce [k

ips]

with tie−bracket without tie−bracket

Figure 2.29 Maximum in-plane force in the bracket for the system with and without intermediate tie-bracket under actual Northridge earthquake.

63

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

5

10

15

20

25

30

35

40

Ratio au / L

For

ce [k

ips]

with tie−bracket without tie−bracket

Figure 2.30 Maximum in-plane force in the bracket for the system with and without intermediate tie-bracket under actual El Centro earthquake.

64

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

20

30

40

50

60

70

Max. Ground Acceleration [g]

Str

ess

[ksi

]

maximum allowable stress

mean mean + one σabsolute maximum

Figure 2.31 Statistics of peak stress in the rail under 50 synthetic ground motions.

65

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

15

20

25

30

35

40

45

50

55

Max. Ground Acceleration [g]

Str

ess

[ksi

]

maximum allowable stress

without tie−bracket with tie−bracket

Figure 2.32 Mean plus one standard deviation of the peak stress in the rail for system with and without intermediate tie-bracket.

66

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 110

−16

10−14

10−12

10−10

10−8

10−6

10−4

10−2

100

Max. Ground Acceleration [g]

Pro

babi

lity

of F

ailu

rewithout tie−bracket with tie−bracket

Figure 2.33 Fragility curves of system with and without intermediate tie-bracket.

67

Table 2.1 Properties of the 10-Story Torsional Building

Mass Inertia

Story [103 lb.s2/in] [105 lb.in.s2]kx = ky [106

lb/in]kt [1012 lb.in] ex [in] ey [in]

1 7.5260 1.4177 4.5855 2.3794 90 60.152 6.9145 1.3025 17.6904 9.7333 90 60.153 6.9145 1.3025 17.6904 9.7333 90 60.154 6.8163 1.2840 13.6469 6.8930 90 60.155 6.8163 1.2840 13.6469 6.8930 90 60.156 6.8163 1.2840 13.6469 6.8930 90 60.157 6.8163 1.2840 13.6469 6.8930 90 60.158 6.6634 1.2550 7.6896 3.7498 90 60.159 6.6634 1.2550 7.6896 3.7498 90 60.1510 6.6634 1.2550 7.6896 3.7498 90 60.15

EccentricityStiffness

68

Table 2.2 Frequencies of the 10-Story Torsional Building

Mode Frequency Period Participation

[rad/sec] [sec] factor1 5.70 1.1014 -139.07852 5.77 1.0892 208.10743 15.60 0.4027 -38.28934 15.78 0.3981 -57.25845 26.33 0.2386 -16.92776 26.63 0.2359 -25.31917 38.59 0.1628 -6.91298 39.04 0.1609 10.35189 47.78 0.1315 -4.753510 48.33 0.1300 7.110611 58.14 0.1081 2.935912 58.81 0.1068 -4.399013 63.23 0.0994 2.266914 63.96 0.0982 -3.377515 73.20 0.0858 -1.666616 74.04 0.0849 2.491217 83.40 0.0753 -0.783418 84.35 0.0745 1.174719 91.33 0.0688 0.974520 92.32 0.0681 -1.454821 967.84 0.0065 0.122922 2605.72 0.0024 0.034723 4400.36 0.0014 -0.015224 6454.48 0.00097 0.005925 8013.25 0.00078 -0.004226 9726.08 0.00065 -0.002427 10585.43 0.00059 0.002428 12336.80 0.00051 -0.001629 14035.72 0.00045 0.000630 15778.94 0.00040 0.0009

69

Table 2.3 Moment of inertia and distance from centroid to outermost point for standard T-sections for guide rail

Rail Size[lb/ft] Ixx Iyy cx cy

8 1.424 1.369 1.641 1.7511 4.279 2.885 2.376 2.2512 4.476 3.948 2.413 2.5015 4.797 6.294 2.544 2.50

18.5 9.743 8.400 2.998 2.7522.5 10.924 9.627 2.652 2.7530 22.633 11.867 3.244 2.75

Moment of Inertia [in4] Distance c [in]

Table 2.4 Seismic inputs used in the study

Event Date Location Max. Ground Acceleration [g] x-direction y-direction El Centro 05/18/1940 Imperial Valley Irrigation District 0.348 0.214 Parkfield 06/27/1966 Cholame, Shandon, CA 0.434 0.355 San Fernando 02/09/1971 Pacoima Dam 1.170 1.075 Whittier 10/01/1987 Tarzana - Cedar Hill Nursery 0.537 0.405 Loma Prieta 10/17/1989 Corralitos - Eureka Canyon Rd. 0.630 0.478 Northridge 01/17/1994 Sylmar - County Hospital Parking Lot 0.843 0.604